System and method for performing object analysis

Abstract

There is provided a system that, when in operation, classifies one or more elements within an input signal, wherein the system: receives a compressed version of the input signal, wherein the compressed version comprises at least two sets of compressed data in a hierarchy, wherein each set of compressed data, when decoded, enables the signal to be reconstructed up to a respective level of quality; decodes the compressed version of the signal up to a first level of quality by decoding a first set of compressed data to generate a first reconstructed signal; performs a first classification operation on the first reconstructed signal; decodes the compressed version of the signal up to a second level of quality by decoding a second set of compressed data to generate a second reconstructed signal; and performs one or more second classification operations on the second reconstructed signal.

The invention claimed is: 
     
       1. A method for using a system to classify one or more elements within an input signal, wherein the method comprises:
 receiving a compressed version of the input signal, wherein the compressed version comprises at least a first set of compressed data and a second set of compressed data in a hierarchy, wherein each set of compressed data, when decoded, enables the signal to be reconstructed up to a respective level of quality; 
 decoding the compressed version of the signal up to a first level of quality by decoding the first set of compressed data to generate a first reconstructed signal; 
 performing a first object classification operation on the first reconstructed signal; 
 decoding the compressed version of the signal up to the second level of quality by decoding a second set of compressed data to generate a second reconstructed signal; and 
 performing one or more second object classification operations on the second reconstructed signal. 
 
     
     
       2. A method of  claim 1 , wherein, when performing the first classification operation, one or more regions of interest in the first reconstructed signal are identified, and wherein the decoding of the compressed version of the signal up to a second level of quality is performed only for those regions of interest. 
     
     
       3. A method of  claim 1 , wherein the first reconstructed signal is at the same resolution as the second reconstructed signal. 
     
     
       4. A method of  claim 1 , wherein the first reconstructed signal is at a mutually different resolution as the second reconstructed signal. 
     
     
       5. A method of  claim 1 , wherein the first classification is performed on a first number of frames in the first reconstructed signal, and the one or more second classifications are performed on a second number of frames in the second reconstructed signal. 
     
     
       6. A method of  claim 1 , wherein the first classification is performed over a full frame in the first reconstructed signal, and the one or more second classifications are performed over a portion of a frame in the second reconstructed signal. 
     
     
       7. A method of  claim 1 , wherein the first classification is performed on a first number of colour planes in the first reconstructed signal, and the one or more second classifications are performed on at least a second number of colour planes in the second reconstructed signal. 
     
     
       8. A method of  claim 7 , wherein the first number of colour planes corresponds to all the planes in the first reconstructed signal, and wherein the second number of colour planes corresponds to a subset of the colour planes in the second reconstructed signal. 
     
     
       9. A method of  claim 7 , wherein the one or more second classifications are performed also on at least a third number of colour planes in the first reconstructed signal, wherein the colour planes included in the second number of colour planes are mutually different from the colour planes included in the third number of colour planes. 
     
     
       10. A method of  claim 9 , wherein the first number of colour planes corresponds to three and includes planes Y, U and V, the second number of colour planes corresponds to 1 and includes the plane Y, and wherein the third number of colour planes corresponds to 2 and includes the colour planes U and V. 
     
     
       11. A method of  claim 1 , wherein the method further comprises:
 decoding the compressed version of the signal up to a third level of quality by decoding a third set of compressed data to generate a third reconstructed signal; and 
 performing one or more third classification operations on the third reconstructed signal. 
 
     
     
       12. A method of  claim 11 , wherein:
 the first classification is performed on a first number of colour planes in the first reconstructed signal, and the one or more second classifications are performed on at least a second number of colour planes in the second reconstructed signal; 
 the one or more second classifications are performed also on at least a third number of colour planes in the first reconstructed signal, 
 wherein the colour planes included in the second number of colour planes are mutually different from the planes included in the third number of colour planes; and 
 the one or more third classifications are performed on at least a fourth number of colour planes in the third reconstructed signal. 
 
     
     
       13. A method of  claim 11 , wherein the third reconstructed signal corresponds to a region of interest of the input signal. 
     
     
       14. A method of  claim 1 , wherein the first level of quality is lower than the second level of quality, wherein quality is defined as a function of image resolution. 
     
     
       15. A method of  claim 1 , wherein the second reconstructed signal is generated partially based also on the first reconstructed signal. 
     
     
       16. A method of  claim 1 , wherein the first classification, one or more second classifications and one or more third classifications are organized in a hierarchical manner, so that the first classification is a coarse classification of some elements in the input signal, and either the one or more second classifications or the one or more third classifications further refine the classification of elements, wherein the method further comprises:
 based on the coarse classification, decoding only those portions of the second set of compressed data which are determined to require a further classification, and generating the second reconstructed signal based on those decoded portions. 
 
     
     
       17. A method of  claim 1 , wherein the classification operation is performed using an artificial intelligence (AI) detector arrangement, wherein the artificial intelligence (AI) detector arrangement includes neural network detectors. 
     
     
       18. A method of  claim 17 , wherein the neural network detectors are implemented such that a lowest neural network layer of the neural network detectors is implemented in at least one FPGA, and one or more higher neural network layers of the neural network detectors are implemented using computing hardware configured to execute a software product. 
     
     
       19. A system that, when in operation, classifies one or more elements within an input signal, wherein the system:
 receives a compressed version of the input signal, wherein the compressed version comprises at least a first set of compressed data and a second set of compressed data in a hierarchy, wherein each set of compressed data, when decoded, enables the signal to be reconstructed up to a respective level of quality; 
 decodes the compressed version of the signal up to the first level of quality by decoding a first set of compressed data to generate a first reconstructed signal; 
 performs a first object classification operation on the first reconstructed signal; 
 decodes the compressed version of the signal up to the second level of quality by decoding a second set of compressed data to generate a second reconstructed signal; and 
 performs one or more second object classification operations on the second reconstructed signal. 
 
     
     
       20. A computer program product comprising a non-transitory computer-readable storage medium having computer-readable instructions stored thereon, the computer-readable instructions being executable by a computerized device comprising processing hardware to:
 receive a compressed version of the input signal, wherein the compressed version comprises at least a first set of compressed data and a second set of compressed data in a hierarchy, wherein each set of compressed data, when decoded, enables the signal to be reconstructed up to a respective level of quality; 
 decode the compressed version of the signal up to the first level of quality by decoding a first set of compressed data to generate a first reconstructed signal; 
 perform a first object classification operation on the first reconstructed signal; 
 decode the compressed version of the signal up to the second level of quality by decoding a second set of compressed data to generate a second reconstructed signal; and 
 perform one or more second object classification operations on the second reconstructed signal.